Adjust how we access thread ID for header thin locks

[jkoritzinsky]

Pass the Thread* object instead of the thread ID and only compute thread ID in the paths that need it to convince the compiler to better separate memory accesses for better pipelining.

This brings the emitted assembly of the .NET 11 implementation closer to the .NET 10 implementation, which I think should be enough to get performance back (as the assembly is darn near identical at this point, arguably better for the new implementation).

[dotnet-policy-service]

Tagging subscribers to this area: @agocke
See info in area-owners.md if you want to be subscribed.

[jkoritzinsky]

@MihuBot benchmark System.Collections.TryAddGiventSize

[Copilot]

Pull request overview

This PR optimizes the thin lock implementation by adjusting how thread IDs are accessed in AcquireHeaderThinLock and ReleaseHeaderThinLock. Instead of passing a pre-computed thread ID (DWORD tid), these methods now accept a Thread* pointer and only call GetThreadId() in code paths where the thread ID is actually needed. This change helps the compiler better separate memory accesses for improved CPU pipelining, bringing the .NET 11 implementation's generated assembly closer to .NET 10's performance characteristics.

Changes:

• Modified function signatures to accept Thread* pCurThread instead of DWORD tid
• Moved GetThreadId() calls inside conditional branches that actually use the thread ID
• Updated call sites to pass GetThread() instead of GetThread()->GetThreadId()

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
src/coreclr/vm/syncblk.h	Updated method signatures for AcquireHeaderThinLock and ReleaseHeaderThinLock to accept Thread* parameter
src/coreclr/vm/syncblk.inl	Modified implementations to defer GetThreadId() calls to only the code paths that need the thread ID value
src/coreclr/vm/comsynchronizable.cpp	Updated call sites to pass GetThread() instead of computing thread ID upfront

[MihuBot]

System.Collections.TryAddGiventSize_String_

BenchmarkDotNet v0.16.0-custom.20260127.101, Linux Ubuntu 24.04.3 LTS (Noble Numbat)
AMD EPYC 9V74 2.60GHz, 1 CPU, 8 logical and 4 physical cores
  Job-POJOIC : .NET 11.0.0 (11.0.0-dev, 42.42.42.42424), X64 RyuJIT x86-64-v4
  Job-TATIAS : .NET 11.0.0 (11.0.0-dev, 42.42.42.42424), X64 RyuJIT x86-64-v4
OutlierMode=Default  PowerPlanMode=  IterationTime=250ms
MaxIterationCount=20  MemoryRandomization=Default  MinIterationCount=15
WarmupCount=1

Method	Toolchain	Count	Mean	Error	Ratio	Allocated	Alloc Ratio
Dictionary	Main	512	6.060 μs	0.0081 μs	1.00	14.38 KB	1.00
Dictionary	PR	512	6.024 μs	0.0090 μs	0.99	14.38 KB	1.00
ConcurrentDictionary	Main	512	20.499 μs	0.0281 μs	1.00	59.21 KB	1.00
ConcurrentDictionary	PR	512	20.536 μs	0.0341 μs	1.00	59.21 KB	1.00